NCAA Rule Change on Tournament Selection Criteria

[Kuiper]

This isn't really related to NPI, per se, but it is about the NCAA Tournament itself, so I thought I would post it here.  basically, the soccer-specific sports committees are exploring the possibility of moving back the championships date a week, which could allow for games to be spread out a bit more with a slightly longer season.  The negative, presumably, is bumping up against review sessions and finals at some schools

https://ncaaorg.s3.amazonaws.com/committees/d3/champs/Jun2024D3CC_Report.pdf

Men's and women's soccer. The committee received an update from the men's and
women's soccer liaisons on the ongoing committee work to survey the membership on a
potential championships date formula adjustment to move the men's and women's soccer
championships one week later than the current formula. . . . The soccer committees are in the process
of receiving . . . membership feedback regarding the potential change to the championship date formula. and will share that once it is analyzed.

[Kuiper]

For those who are really interested in almost 1.5 hours of NPI, Here is a link to the recording of the webinar NCAA's assistant director of championships conducted for explaining the new NPI NCAA tournament selection system for men's soccer.

I have seen more and more schools have one assistant coach who if not formerly called the director of sports analytics, effectively is assigned that role.  For schools focused on maximizing their chances under the NPI system, I expect that analytics assistant coach will likely have a much bigger role in scheduling decisions and in-game decisions about the value of a win v. a tie to take into account the new NPI system.

[Kuiper]

Here is an interesting pro-NPI point from a D3 hoops perspective that I assume will apply to D3 soccer too.  I think The basic insight is that the old "wins over regionally ranked opponents" component of the selection process didn't differentiate much if at all between regions in giving credit for a win over a regionally ranked opponent. Presumably, NPI gives more of a bump for wins against top opponents in a national ranking than wins against a top 10 opponent from a weak region.

https://x.com/iwuhoopscom/status/1817925477572788580?s=46&t=uXiupHZfR0TxrRyWF9BmRg

"In learning more about NPI from the episode, it seems NPI will do a better job giving special credit for "good wins" than the RRO system did.

RROs had good intention but was extremely flawed...mainly because of differing strengths of regions."

[Kuiper]

The basketball guys at D3 Datacast did a whole show and supplemental Q&A on the new NPI system and how it will work.  Some of it is of general application and may be of interest:

https://d3datacast.com/2024/08/12/npi-qa-episode-83-supplement/

There was a specific question about ties in soccer that they added to the Q&A

Question 5: In soccer, how will draws work? I would assume they count as half a win and half a loss. Thus, are you able to drop just one side of the draw if the win part hurts you or the loss part helps you? Or do you have to drop the entire match? Or can you even drop a draw?

Answer 5: You are correct that ties are counted as half a win and half a loss. However, we believe NPI does not treat these as separate inputs so there is no dropping one side or the other –  it goes in as one Game NPI. The way it would work in the calculation we discussed at about the 24:17 mark of our video is that the winning percentage component would be 50 and the multiplier on any QWB would be cut in half. Or less directly, the Game NPI could be computed had it been a full win or a full loss then take the average but the result is the same. Our current understanding is that ties do not get dropped.

[ziggy]

Hey, Zac from the D3 Datacast here. We need to correct the answer about ties we gave earlier today. With the current dial settings it may be a narrow scenario in which it would happen, but it is technically possible for the win component of a tie to be dropped from the NPI calculation.

We have updated the response on our site but I wanted to also include the corrected version here directly:

Ties are counted as half a win and half a loss. These two halves of the result — the win half and the loss half — are essentially handled separately. If the "win half" of the result is one that would hurt NPI (barring minimum wins threshold), it can be dropped from the calculation.

As a (hopefully) clarifying example, one win and one loss versus the same opponent would result in the same NPI as two ties versus that opponent. If you would drop the 'true' win versus that team, you would also drop the two 'half wins' from the ties.

[Crossit4fun]

This was a response from the NCAA where I asked about meeting the 10 win minimum...regarding ties in the NPI..

One note, specific to soccer, with ties – those equal half wins. So a team that is 8-2-4 would keep the eight wins, and then four ties is equal to four half wins or two wins an that would equal 10. So even though it is listed as wins, ties do play a roll in hitting that minimum for those looking to drop a win against a bad opponent that hurts its NPI.

[SimpleCoach]

Hey, Zac from the D3 Datacast here. We need to correct the answer about ties we gave earlier today. With the current dial settings it may be a narrow scenario in which it would happen, but it is technically possible for the win component of a tie to be dropped from the NPI calculation.

We have updated the response on our site but I wanted to also include the corrected version here directly:

Ties are counted as half a win and half a loss. These two halves of the result — the win half and the loss half — are essentially handled separately. If the "win half" of the result is one that would hurt NPI (barring minimum wins threshold), it can be dropped from the calculation.

As a (hopefully) clarifying example, one win and one loss versus the same opponent would result in the same NPI as two ties versus that opponent. If you would drop the 'true' win versus that team, you would also drop the two 'half wins' from the ties.

Thanks for the clarification @Zac.

Am I the only one who finds this unnecessarily complicated?  Seems there are some tried and true ways to "rank" a team base on their W-L-T record, and here we have a system that obfuscates all that.

I confess, my opinion may be shaped from ignorance, so I am happy to have my mind changed.  But right now it has me scratching my head.

SC.

[ziggy]

This was a response from the NCAA where I asked about meeting the 10 win minimum...regarding ties in the NPI..

One note, specific to soccer, with ties – those equal half wins. So a team that is 8-2-4 would keep the eight wins, and then four ties is equal to four half wins or two wins an that would equal 10. So even though it is listed as wins, ties do play a roll in hitting that minimum for those looking to drop a win against a bad opponent that hurts its NPI.

Yes, this part was well-understood but got trickier trying to understand if one of those "half results" could be dropped by NPI in the same way their "full result" counterparts can be. Beyond the minimum wins threshold, wins that would drop your NPI are excluded.

If your next-best win above the minimum win dial would drop your NPI it is excluded so that you are not hurt by it. Should that "win" be a half win from a tie, it sets up a situation where only part of a game's results ends up on a team's resume - the loss would be on the resume but the win portion could be dropped from the calculation. It seems a little weird but perhaps makes more sense when you consider that two ties against a team are seen as equal to going 1-1 against that same team and potentially being able to drop two half wins in the 0-0-2 scenario is the same as being able to drop the win in the 1-1 scenario.

Now here is the thing: While NPI doesn't prevent this, I'm not sure how realistic it is that it would actually happen. It would take a combination of ten wins against really good competition and a tie against a really weak opponent. It is possible but my gut says that the minimum wins dial is set high enough that it is unlikely.

Hopefully this is helpful and not too "in the weeds" 

[Kuiper]

Some updates on changes to selection and bracketing criteria for the NCAA DIII tournament in the Summary of Summer 2024 Actions taking by the Presidents and Management Councils.  The first is most significant potentially, if it means we avoid the groups of death because of the extreme geographical concentration of DIII schools in certain parts of the country.

1.  The Management Council approved a two year waiver for the 500 mile limit on travel so that top seeds don't have to face each other in the early rounds:

NCAA Bylaw 31.3.5.1 (b) – Executive Regulations – Selection of
Teams and Individuals for Championships Participation –
Selection Decisions of Sport Committees – Pairings and Site
Selection – Protecting Top-Seeded Teams.

Management Council. The council approved a waiver of Bylaw 31.3.5.1 (b) that requires geographic proximity (i.e., the competition site is within 500 miles of the institution) to take precedent over seeding when constructing a championship bracket. The waiver allows flexibility for sport committees to ensure that top seeded teams do not face each other in the early rounds of championships. The waiver applies for up to two years to allow time
for the impact of the change to be implemented and assessed.

Presidents Council. No action was necessary.

2. Management Council approved an administrative regulation that codifies existing practice that conference opponents will not meet in the first round of the tournament and that seeding will be considered, but not be determinative in who hosts.

NCAA Bylaw 31.3.5.1 (b) – Executive Regulations – Selection of
Teams and Individuals for Championships Participation –
Selection Decisions of Sport Committees – Pairings and Site
Selection – Geographic Proximity.

Management Council. The council approved, in concept, an administrative regulation that amends Bylaw 31.3.5.1 to (a) codify the existing practice that conference opponents will not meet in the first round of championships; and (b) that seeding will be considered but not be determinative when selecting hosts for championships.

Presidents Council. No action was necessary.

[Crossit4fun]

Does anyone know when the first NPI ranking will be published?

Assuming around the second Tuesday/Wed. in Sept. consistent with prior years rankings being published

[ts33]

Does anyone know when the first NPI ranking will be published?

Assuming around the second Tuesday/Wed. in Sept. consistent with prior years rankings being published

Where can we find this once posted? Will this include all teams at the d3 level? or only the specific number of teams necessary to fill the bracket (prior to aq's)?

[Crossit4fun]

In the NPI FAQ's it stated it would published on the NCAA website landing page, no link or url was provided

They also wrote the regional rankings of the top 7 teams would still be published but not used its informational only as NPI is now the tool used for rankings

[Crossit4fun]

From the NCAA (I had asked the date of first NPI release and where the NPI link would be located):

DIII champs committee is still finalizing when the first rank will be released. It more than likely will be around the time of when our first mock ranking would have taken place – which is around October 16, but we still have to see on that becoming official.
 
The plan for the link will be to have it available through NCAA stats as well as a link on the rankings page as well when it all becomes available. The new system will be shown through NCAA stats software and so it will look a little different than what you may have seen previously. So in this first year it will be a little more limited than originally planned as they want to make sure everything is working properly and that the results are all entered correctly before advancing the system. They don't want to go full out and have to pull back, so it will be an on going improvement process over the first year or two starting with the fall.

[College Soccer Observer]

Interesting to see how this has all played out so far.  All 8 NESCAC teams that made the conference tournament were in the top 34 as of 10/30.  Midd 1, Tufts 3, Amherst 4, Williams 8, Conn 10, Wesleyan 16, Bowdoin 33, Hamilton 34.
NCAC has Kenyon at 9, Ohio Wesleyan 15, and Denison 18
Centennial has Dickinson 12, Hopkins 14, F&M 21, Muhlenberg 37
MIAC has Gustavus Adolphus 5, St. Olaf 24, Macalester 29

The system may end up with MORE bids for the NESCAC that happened under the old system. 

[Kuiper]

Interesting to see how this has all played out so far.  All 8 NESCAC teams that made the conference tournament were in the top 34 as of 10/30.  Midd 1, Tufts 3, Amherst 4, Williams 8, Conn 10, Wesleyan 16, Bowdoin 33, Hamilton 34.
NCAC has Kenyon at 9, Ohio Wesleyan 15, and Denison 18
Centennial has Dickinson 12, Hopkins 14, F&M 21, Muhlenberg 37
MIAC has Gustavus Adolphus 5, St. Olaf 24, Macalester 29

The system may end up with MORE bids for the NESCAC that happened under the old system. 

You probably have to distinguish the NPI structure in its bare conceptual form and the one that exists after the Men's Soccer Committee set the dials.  I think the 10 win minimum to drop bad wins and the QWB advantage really changed things from the skeletal version, which accounts for some of the advantages for strong conferences.